Objective: Cushing's disease (CD) may recur despite corticotropic insufficiency (COI) following pituitary surgery. The predictive value of the desmopressin test (DT) for recurrence in this setting remains controversial. We have evaluated whether the disappearance of the response to DT predicts a low probability recurrence in a large cohort of patients with post-operative COI. Design: Multicentre retrospective study. Methods: Ninety-five patients with CD (women 82%, age 41 ± 14 years), responding preoperatively to DT and with early post-operative COI (08 00 am cortisol: <138 nmol/L), underwent a DT within 3 months post-surgery. Association between DT findings and the prediction of recurrence was tested using regression and ROC analyses. Results: Recurrence occurred in 17/95 patients within 29 to 91 months. The cortisol peak (327, 95% CI (237-417) vs 121 (79-164) nmol/L, P = 0.0001) and absolute increment during DT (208 (136-280) vs 56 (22-90) nmol/L, P = 0.005) were greater in the recurrence vs remission group. Cortisol peak (AUC: 0.786 (0.670-0.902)) and increment (0.793 (0.672- 0.914)) yielded a higher prognostic performance for recurrence than did the early post-operative 08 00 am cortisol (0.655 (0.505-0.804)). In the context of COI, cortisol peak >100 nmol/L and increment >30 nmol/L had a high negative predictive value (94, 95% CI (88-100) and 94, (88-100), respectively). Patients with a cortisol peak ≤100 nmol/L (vs >100) or an increment ≤30 nmol/L (vs >30) were less likely to have CD recurrence (odds ratios: 0.12, 95% CI (0.03-0.41) and 0.11 (0.02-0.36), respectively). Conclusion: The disappearance of the response to the post-operative DT was independently associated with a lower odds of CD recurrence and offers an incremental prognostic value, which may help to stratify patients with COI and refine their follow-up according to the risk of recurrence. © 2020 European Society of Endocrinology.< Réduire